The TEF approach for hexachlorobenzene.

same variety of soybeans grown in different locations (25,26). Also, in closed formula diets containing soybeans, the concentration of soybeans in the diet is not reported and may vary from batch to batch depending upon the supply and demand of dietary ingredients. This could explain why the phytoestrogen content of diets may vary greatly from batch to batch or from the same diet processed by different rodent diet vendors. These findings support the need to use open formula diets prepared from controlled ingredients and the need to monitor these diets for estrogenic substances that affect biological end points. We also agree with Boettger-Tong et al. (1) that rodent diet vendors do not routinely monitor all diets for estrogenic substances including phytoestrogens. Rodent diet vendors producing certified diets for use in comparative estrogenicity and carcinogenicity studies should provide the user with a list of substances assayed, including phytoestrogens, and the results. In our opinion, diets used in comparative estrogenicity or carcinogenicity studies should contain nondetectable levels of estrogenic substances that may alter research results. These diets should be monitored for estrogenic substances and their concentrations reported. The commentary by Boettger-Tong et al. (1) is important because the authors emphasize the important role of the animal's diet, especially phytoestrogens, when conducting animal bioassays for estrogenicity or studies that are influenced by in vivo end points of hormone action. We previously reported that rodent diets significantly differ in estrogenic activity and that a standardized diet with minimal estrogenic activity would be desirable for comparative bioassays for estrogenic substances (5). Results from our second study (20) confirm that a standardized open formula diet should be used for studies that are influenced by exogenous estrogens (4,5). Phytoestrogens were not detected in the AIN-76A and the AIN-93M purified casein diets. Therefore, careful consideration should be given to the use of diets such as these when conducting studies that are influenced by exogenous estrogens. We have also shown that a natural ingredient diet can be formulated to contain less than detectable levels of the phytoestrogens (daidzein and genistein) by omitting soybean and alfalfa meals. The soybean and alfalfa meals were omitted because they may be a source of multiple, yet unidentified, phytoestrogens that may further complicate the interpretation of results from studies which are influenced by exogenous estrogens. We recommend that studies to determine the effects of dietary phytoestrogens on results of toxicloOgiC investigations are important and timely. National Toxicology Program studies that will help us understand the effect of long-term feeding of a diet with less than detectable levels of daidzein and genistein are presendy under way. In conclusion, we recommend that careful consideration be given to the phytoestrogen content of the diet when conducting studies that are influenced by exogenous estrogens. A standardized open formula diet in which estrogenic substances have been reduced to minimal levels or to less than detectable levels of daidzein and genistein is recommended for use in such studies. In addition, the selected diet should be monitored for estrogenic substances, and their concentrations should be reported.

same variety of soybeans grown in different locations (25,26). Also, in closed formula diets containing soybeans, the concentration of soybeans in the diet is not reported and may vary from batch to batch depending upon the supply and demand of dietary ingredients. This could explain why the phytoestrogen content of diets may vary greatly from batch to batch or from the same diet processed by different rodent diet vendors. These findings support the need to use open formula diets prepared from controlled ingredients and the need to monitor these diets for estrogenic substances that affect biological end points. We also agree with Boettger-Tong et al. (1) that rodent diet vendors do not routinely monitor all diets for estrogenic substances including phytoestrogens. Rodent diet vendors producing certified diets for use in comparative estrogenicity and carcinogenicity studies should provide the user with a list of substances assayed, including phytoestrogens, and the results. In our opinion, diets used in comparative estrogenicity or carcinogenicity studies should contain nondetectable levels of estrogenic substances that may alter research results. These diets should be monitored for estrogenic substances and their concentrations reported.
The commentary by Boettger-Tong et al. (1) is important because the authors emphasize the important role of the animal's diet, especially phytoestrogens, when conducting animal bioassays for estrogenicity or studies that are influenced by in vivo end points of hormone action. We previously reported that rodent diets significantly differ in estrogenic activity and that a standardized diet with minimal estrogenic activity would be desirable for comparative bioassays for estrogenic substances (5). Results from our second study (20) confirm that a standardized open formula diet should be used for studies that are influenced by exogenous estrogens (4,5). Phytoestrogens were not detected in the AIN-76A and the AIN-93M purified casein diets. Therefore, careful consideration should be given to the use of diets such as these when conducting studies that are influenced by exogenous estrogens. We have also shown that a natural ingredient diet can be formulated to contain less than detectable levels of the phytoestrogens (daidzein and genistein) by omitting soybean and alfalfa meals. The soybean and alfalfa meals were omitted because they may be a source of multiple, yet unidentified, phytoestrogens that may further complicate the interpretation of results from studies which are influenced by exogenous estrogens. We recommend that studies to determine the effects of dietary phytoestrogens on results of toxicloOgiC investigations are important and timely. National Toxicology Program studies that will help us understand the effect of long-term feeding of a diet with less than detectable levels of daidzein and genistein are presendy under way.
In conclusion, we recommend that careful consideration be given to the phytoestrogen content of the diet when conducting studies that are influenced by exogenous estrogens. A standardized open formula diet in which estrogenic substances have been reduced to minimal levels or to less than detectable levels of daidzein and genistein is recommended for use in such studies. In addition, the selected diet should be monitored for estrogenic substances, and their concentrations should be reported.

Correspondence
There is no PF for TCDD currently available in the Integrated Risk Information Service database (3). However, the PF originally derived from the aforementioned study was 156,000 mg/kg/day and may be modified downward to approximately 100,000 mg/kg/day, based on reinterpretation of histopathologic data. Applying van Birgelen's TEF to these values would suggest a PF for the carcinogenic action of HCB (under the linearized multistage assumptions) to be approximately 10-16 mg/kg/day, or about an order of magnitude more potent than is suggested by the in vivo data. It is possible that the difference lies in pharmacokinetic or pharmacodynamic factors that exist in the whole animal, which cannot be captured in cellular assays.
As van Birgelen points out (1), the World Health Organization indicates a preference for long-term in vivo studies over in vitro measures when setting a TEF (5). The HCB example illustrates the extreme caution that should be exercised when applying a TEF based solely on in vitro information, particularly when major public health claims are being made.

Bradley W. Schwab Ogden Environmental and Energy
Services Company Westford, Massachusetts Response: Hexachlorobenzene I appreciate Schwab's comments regarding caution in the use of a toxic equivalency factor (TEF) for hexachlorobenzene (HCB) based on results of in vitro studies. The dioxinlike effects of HCB include cytochrome P4501A induction and binding to the aryl hydrocarbon (Ah) receptor. In addition, HCB has been shown to bioaccumulate. These three factors are a prerequisite to include a compound in the TEF concept, which compares the potency of a dioxinlike compound to 2,3,7,8-tetrachlorodibenzo-p-dioxin (TCDD). TEFs are consensus values based on available data on relative potency values for specific compounds (1). TEF values are used to estimate the total dioxin activity in environmental and human samples by multiplying the TEF value by the concentration of each compound, leading to a certain amount of toxic equivalents (TEQs) for each compound. The summation of all TEQs in a certain mixture expresses the total dioxin activity of this mixture. Based on the binding affinity of HCB to the Ah receptor, in vitro cytochrome P4501A induction, and porphyrin accumulation, a relative potency of 0.0001 for HCB was estimated (X). Using this relative potency value suggested that HCB could lead to a considerable contribution to the dioxin activity of human milk in some countries. I did not estimate the slope factor for HCB that is used in carcinogenicity assessment. The slope factor is the result of the application of a low-dose extrapolation procedure and is presented as the risk per milligram per kilogram of body weight per day (mglkg/day) (3).
Schwab's comments included the comparison of these slope factors (although potency factors are mentioned) for HCB (1.6 per mg/kg/day) and TCDD (100,000 or 156,000 per mglkg/day). The slope factors for TCDD are not available in the Integrated Risk Information Service database (3), as Schwab mentioned. He points out correctly that the ratio between these two slope factors is different from the suggested relative potency value for HCB. He assumes by using this approach that TEFs would predict the carcinogenic potential of dioxinlike compounds. However, no studies have been performed to verify this approach. Studies are currently under way to determine whether relative potency values based on biochemical effects are predictive for carcinogenesis in female Sprague Dawley rats for various dioxinlike compounds (4).